Oil field operators dedicate significant resources to improve the recovery of hydrocarbons from reservoirs while reducing recovery costs. To achieve these goals, reservoir engineers both monitor the current state of the reservoir and attempt to predict future behavior given a set of current and/or postulated conditions. Reservoir monitoring, is sometimes referred to as reservoir surveillance, involves the regular collection and monitoring of measured production data from within and around the wells of a reservoir. Such data may include, but is not limited to, water saturation, water and oil cuts, fluid pressure and fluid flow rates. As the data is collected, it is archived into a historical database.
The collected production data, however, mostly reflects conditions immediately around the reservoir wells. To provide a more complete picture of the state of a reservoir, simulations are executed that model the overall behavior of the entire reservoir based on the collected data, both current and historical. These simulations estimate the reservoir's overall current state, producing simulated interwell data values both near and at a distance from the wellbores. Simulated near-wellbore data is regularly correlated against measured near-wellbore data, with modeling parameters being adjusted as needed to reduce the error between the simulated and measured data. Once so adjusted, the simulated interwell data, both near and at a distance from the wellbore, may be relied upon to assess the overall state of the reservoir. Such data may also be relied upon to predict the future behavior of the reservoir based upon either actual or hypothetical conditions input by an operator of the simulator. The results of such predictive simulations may subsequently be used to determine optimal parameters for operating the wells within the reservoirs and thus maximize reservoir production.
Reservoir simulations, particularly those that perform full physics numerical simulations of large reservoirs, are computationally intensive and can take hours, even days to execute. To reduce the execution time and cost of these simulations, individual reservoir fluid components, which can number in the hundreds, are sometimes grouped or “lumped” together into pseudo-components. Each pseudo-component is treated by the simulation as a single component. The components are selected for lumping into a pseudo-component based on a common characteristic within a given range (e.g., molar mass).
For large fields with multiple reservoirs, a gathering network of pipelines and other equipment coupled to the reservoirs collects the extracted product and transports it downstream for further processing. As with the reservoirs, the gathering network is also monitored and simulated to predict how changes in reservoir conditions affect and/or will affect the product provided to a downstream facility. However, as already noted the full simulation of all components of just one reservoir can be prohibitively expensive and time-consuming. Thus, fully simulating multiple reservoirs together with the gathering network to determine the mixed fluids' behavior within the gathering network is generally considered impractical.
Similarly, while using pseudo-components may reduce the number of simulated reservoir components, each reservoir's components are typically different and in different proportions, resulting in different pseudo-components for each reservoir and thus in many cases a prohibitively large number of pseudo-components to simulate together with the gathering network. To overcome this, each reservoir may be simulated separately for a short period of time, with the results of these simulations being used as boundary conditions for a gathering network simulation. The results of the gathering network simulation are then used as boundary conditions for the reservoir simulation for a subsequent time period. The sequence of alternately performing a reservoir simulation followed by a network simulation is repeated until the whole time period of interest has been simulated, and is sometimes referred to as loose coupling. Over time, however, this approach can result in large errors in the simulation results, and in some cases the simulation results may exhibit large oscillations. To help reduce such errors, multiple iterations may be attempted over each time period, an approach sometimes referred to as iterative coupling. Such an approach, however, can be costly and may fail to converge at all.
It should be understood that the drawings and corresponding detailed description do not limit the disclosure, but on the contrary, they provide the foundation for understanding all modifications, equivalents, and alternatives falling within the scope of the appended claims.